Morning Briefing | Apple Reportedly Uses Google AI to Train Its Own Models / Huang Renxun: I’ll Explode If Engineers Don’t Burn Tokens / MIIT: EV Batteries Enter Mass-Scale Retirement Phase

🤖

Apple Reportedly Trains On-Device AI with Gemini; Some Siri Requests Routed to Google Cloud

🚗

Seres Responds to Tesla’s FSD Entry into China

🧠

Anthropic Prioritizes Values-Based Interviews in Hiring Process

🔍

ChatGPT’s Consumer AI Traffic Share Narrowed; Market Enters “Three-Power Balance”

💡

Huang Renxun on AI Spending: “Wasting Money Is Fine—Don’t Waste Time”

📋

Liu Qiangdong Proposes “Three 12s” Principle; JD Caps Daily Work Hours at 12

☁️

$500M Burned in One Month: Corporate AI Bills Spin Out of Control

⛽

MIIT: Power Batteries Enter Large-Scale Retirement Phase

🦿

Unitree’s Embodied Intelligence Experience Center—Asia’s First Store—to Open in Shanghai Jing’an

🖨️

BMW to Deploy Humanoid Robots at Leipzig Plant

🚙

Hello Ride Clarifies Expressway Toll Rules for Ride-Sharing

💰

Zhipu’s Market Cap Briefly Reaches HK$880 Billion

💡

Mo Yan: AI Creation Is Essentially “Secondhand Goods”—Cannot Replace Writers’ Originality

📱

Starting at ¥2,899, vivo S60 Series Launched

📱

Xiaomi 17T Series to Launch in Mainland China

🎨

Base Kit at ¥8,999: Haifu AQUA Launched—Zero Zero Tech Brings Flying Cameras to Water Sports

📱

Starting at ¥2,699, Honor WIN Turbo Launched

📰 Weekend Reads Worth Your Attention

Apple Reportedly Trains On-Device AI with Gemini; Some Siri Requests Routed to Google Cloud

According to *The Information*, Apple plans to significantly ramp up on-device AI computing power at next month’s Worldwide Developers Conference (WWDC), while integrating NVIDIA’s privacy-preserving technologies to secure certain cloud-based AI tasks.

Sources reveal that Apple is using Google’s large language model Gemini to train lightweight models capable of running locally—a process known as model distillation.

To accelerate model compression, Apple is exploring acquisitions of relevant startups. Liquid AI, an edge-AI company based in Cambridge, Massachusetts, was reportedly among potential acquisition targets.

For complex cloud-based requests beyond the capabilities of on-device processing, Apple has approved the use of NVIDIA’s Confidential Compute technology on Google Cloud. This hardware-level encryption ensures data and models remain protected during processing, allowing Apple to uphold its “Private Cloud Compute” branding while still relying on NVIDIA GPUs and Google’s cloud infrastructure behind the scenes.

Seres Responds to Tesla FSD Entry into China

According to IT Home, at the recent Fourth Future Automotive Pioneers Forum, Zhang Zhengping, Chairman of Seres Automobile, shared his views comparing Huawei’s Qiankun Intelligent Driving system with Tesla’s Full Self-Driving (FSD).

He stated that the superiority of either system should ultimately be judged objectively by the market. Regarding technical approaches, Zhang noted:

“Which intelligent driving system performs better must be answered by the market. The all-new AITO M9 is equipped with six LiDAR units—this matches or exceeds FSD’s capability. Moreover, FSD mimics human visual driving logic, whereas Huawei’s Qiankun Intelligent Driving achieves a state where it ‘sees beyond what human eyes can perceive.’”

Tesla’s FSD, since version V12, has adopted a pure-vision approach, leveraging eight cameras and an end-to-end neural network architecture centered on simulating human visual reasoning.

Anthropic Prioritizes Values-Based Interviews in Hiring Process

Bloomberg Businessweek reported on Anthropic’s hiring practices, noting that despite rapid expansion, the AI firm places candidate alignment with its mission at the core of its interview process; applicants typically undergo multiple rounds of skills assessments and interviews, with most exercises requiring completion without AI assistance.

The report highlights that Anthropic’s culture interviews probe candidates’ values, worldviews, and perspectives on AI risks. Multiple candidates and career coaches described this round as deeper than typical corporate culture interviews—even delving into ethical dilemmas, how candidates responded at the time, and their retrospective reflections.

Anthropic President Daniela Amodei previously mentioned on a podcast that the company asks candidates about beliefs they hold that are “unusual but worth defending,” focusing particularly on how individuals uphold their judgments in uncomfortable situations.

ChatGPT Consumer AI Traffic Share Narrowed; Market Enters “Three-Power Balance”

Latest traffic data from SimilarWeb shows the consumer AI market shifting from ChatGPT’s dominance toward a “three-power balance” among OpenAI, Google, and Anthropic.

Over the past six months, ChatGPT’s share dropped from ~80% to ~60%, while Gemini’s relative traffic rose from ~20% to ~50%, and Claude increased from ~3% to ~20%.

Despite this shift, ChatGPT remains the leader. Dongcha notes that ChatGPT has ~900 million weekly active users, Gemini ~500 million, and Claude ~150–200 million; Claude recorded the fastest quarterly growth among top websites, surging 255%.

Huang Renxun on AI Spending: “Wasting Money Is Fine—Don’t Waste Time”

According to 36Kr, NVIDIA CEO Huang Renxun recently addressed concerns over AI usage costs during an internal company meeting, stating: “It’s fine to waste a little money—but never waste time.”

During the meeting, employees expressed concern that current heavy AI usage merely burns tokens without delivering tangible productivity gains. Huang responded that, as with any emerging technology, mastery takes time, so perfection shouldn’t be expected upfront—instead, teams should take the first step and embrace new tools.

On AI cost management, Huang consistently advocates aggressive AI adoption. He previously remarked that for an engineer earning $500,000 annually, he’d worry if their token expenses fell below $250,000—and would be “furious” if they spent only $5,000.

His rationale: spending half an employee’s salary on AI is justified if it yields a tenfold efficiency gain.

Liu Qiangdong Proposes “Three 12s” Principle; JD Caps Daily Work Hours at 12

According to TechWeb, JD Group recently introduced a new work-hour management policy called the “Three 12s,” proposed by founder Liu Qiangdong last week and now being rolled out across multiple departments. Notifications have already been disseminated from C1/C2 leadership levels downward, with frontline R&D teams receiving specific implementation guidance.

The “Three 12s” principle centers on three quantifiable metrics:

• First, daily working hours should not exceed 12 hours, except during major promotional periods or special projects;
• Second, employees are encouraged to leave the office no later than midnight, and all offices must guide on-site staff to depart and switch off lights by 24:00;
• Third, continuous on-duty work days must not exceed 12 to prevent chronic overtime without rest.

JD stated this policy is a proactive adjustment targeting excessive workloads in some internal teams, explicitly incorporating employee physical and mental well-being into management priorities through quantified work-hour standards to regulate daily operations and overtime practices.

$500M Burned in One Month: Corporate AI Bills Spin Out of Control

Per Axios, enterprises are increasingly questioning rising AI expenditures, with cost control and ROI becoming central tensions in 2026 corporate AI procurement decisions.

Microsoft recently canceled most of its Claude Code licenses, primarily due to cost concerns; Uber’s COO publicly stated AI-related spending is “becoming harder to justify.”

An AI consultant told Axios that one client, having failed to cap employee usage of Claude licenses, burned $500 million in a single month.

CloudBees CEO Anuj Kapur told Axios that some companies are citing AI automation to justify layoffs—though layoffs may simply be “the only lever they can pull” to offset soaring AI bills. Meanwhile, consumer sentiment toward AI is cooling, and employee resistance is also growing.

MIIT: Power Batteries Enter Large-Scale Retirement Phase

On May 28, the Ministry of Industry and Information Technology (MIIT) held the second meeting of the National Working Group on New Energy Vehicle Power Battery Recycling and Utilization, declaring that China’s EV power batteries have entered the large-scale retirement phase. Accelerating the establishment of a robust recycling and reuse system has become a key priority for 2026.

The meeting emphasized coordinated efforts in legislation, policy, standard-setting, and critical research, alongside launching joint enforcement actions targeting illegal battery recycling practices—including unauthorized resale of retired batteries, manufacturing substandard products from used batteries, failure to fulfill information traceability obligations, illegal dismantling causing environmental pollution, and unlicensed operations.

MIIT also urged enterprises to strengthen primary responsibility, leverage digital technologies to monitor battery flows, and encourage collaboration among industry associations, key enterprises, and research institutions to explore advanced equipment and novel business models.

Unitree’s Embodied Intelligence Experience Center—Asia’s First Store—to Open in Shanghai Jing’an

Unitree Technology announced that Asia’s first Unitree Embodied Intelligence Experience Center will open tomorrow in Shanghai. Located on the 2nd floor of Jiuguang Department Store at 1618 Nanjing West Road, Jing’an District, the store spans over 100 square meters—directly across from Apple Jing’an—and will showcase humanoid robots and quadruped robot dogs.

BMW to Deploy Humanoid Robots at Leipzig Plant

According to BBC, BMW plans to deploy humanoid robots in European automotive production this summer. Two Aeon robots, manufactured by Hexagon Robotics, are currently undergoing testing at the Leipzig plant. Michael Nikolaides, BMW’s Head of Process Management and Digitalization, stated that humanoid robots represent the future of automobile manufacturing.

The Aeon stands 1.65 meters tall, weighs 60 kg, reaches a top speed of 2.4 m/s, can carry 15 kg briefly, and sustainably lifts 8 kg. It features 21 sensors—including cameras, radar, microphones, force sensors, and torque sensors—and operates for approximately 3 hours per battery charge before requiring replacement.

Hello Ride Clarifies Expressway Toll Rules for Ride-Sharing

Hello Ride released guidelines yesterday clarifying expressway toll rules for ride-sharing. The core principle: toll allocation follows the passenger’s note; driver acceptance of the ride implies agreement with the specified arrangement.

• If the passenger selects “willing to negotiate fees,” both parties must reach consensus via the platform chat or phone call; drivers must finalize negotiations before accepting the ride or within 2 hours after acceptance—and before the passenger boards; otherwise, tolls default to non-charged.
• If the passenger selects “I’ll cover all fees,” they bear the full expressway toll for the route between pickup and drop-off points.
• If the passenger selects “I won’t pay,” they bear no tolls, and the driver decides whether to take the expressway.

The guidelines also list violations: charging passengers who declined toll payments, duplicate toll charges, charging without actual toll incurred, charging amounts inconsistent with the trip, or refusing payment after prior agreement.

Penalties include refunding overcharged amounts, deduction of growth points and trust scores, 3–7 day suspensions, or permanent bans for severe cases; passengers failing to pay agreed tolls will be blocked from booking until payment is completed. Hello Ride reminds users to pay tolls exclusively via the ride-sharing platform—not offline transfers.

MiniMax Officially Launches A-Share IPO Process

Per the China Securities Regulatory Commission’s official website, MiniMax signed a tutoring agreement with CITIC Securities on May 29, officially initiating its A-share IPO process.

Earlier, Zhipu had signed a tutoring agreement with Guotai Junan on February 9 this year and is jointly conducting STAR Market tutoring with previously engaged CICC. MiniMax’s launch signals that the “two titans of large models”—Zhipu and MiniMax—may go public on the A-share market simultaneously.

Zhipu’s Market Cap Reaches HK$880 Billion

According to Caixin, Zhipu AI’s Hong Kong-listed market cap briefly surged to approximately HK$880 billion, nearly 17 times its January IPO price of HK$116.2 per share. As of yesterday’s close, its market cap stood at HK$711.12 billion.

💡 Mo Yan: AI Creation Is Essentially “Secondhand Goods”—Cannot Replace Writers’ Originality

In a recent Xinhua interview, Mo Yan discussed AI and literary creation, asserting that AI cannot replace writers.

“I believe the most precious asset a writer possesses is originality—the ability to write novels or poems never written before, by oneself or anyone else, and to create archetypal characters unseen in others’ works. That is the very reason and value for a writer’s existence.”

Mo Yan further explained that current AI-generated content lacks true originality:

“AI is trained on texts written by generations of authors. Based on this vast corpus, it recombines elements to produce seemingly novel works. But ultimately, it’s not original—it remains secondhand.”

Mo Yan also acknowledged that the application of AI as an auxiliary tool has become a trend, particularly in translation work, where its value will gradually become evident; in literary planning, film and television production, and technical aspects, AI will grow increasingly important, potentially disrupting certain job roles.

During the interview, he further discussed AI’s divisive effect on people: tools may empower intelligent individuals to become even sharper, while those lacking judgment might grow more reliant on pre-generated outputs.

vivo S60 Series Launched Starting at CNY 2,899

The vivo S60 series was officially launched yesterday, with the vivo S60 starting at CNY 3,599 and the vivo S60 Yuanqi Edition starting at CNY 2,899. The series is endorsed by actress Tian Xiwēi and comes in three color options: “Starry Sea,” “Early Summer Green,” and “Midsummer Night.” The device measures 7.92 mm thick and weighs 199 g.

• Display: 6.59-inch 1260p 144Hz flat screen
• Imaging: 50MP front camera; S60 features a 50MP large-sensor main camera, 8MP ultra-wide, and 50MP IMX882 3X periscope telephoto; S60 Yuanqi Edition includes a 50MP large-sensor main camera and 8MP ultra-wide
• Performance: S60 powered by Snapdragon 8s Gen 3; Yuanqi Edition powered by Dimensity 7500
• Battery: Both models feature a 7,200mAh Blue Ocean Battery with 90W wired charging
• Other: Standard 3D ultrasonic fingerprint sensor, IP68/IP69 rating, stereo dual speakers, IR remote control, and X-axis linear motor

vivo emphasized 4K Live imaging and Xiaohongshu publishing scenarios. Officially, the S60 supports 4K Live, Starlight Live, Live Color Palette, Inspiration Sticker Live, and generates 3D spatial effects from Live cover frames; vivo claims the S60 is the industry’s only device supporting 3D spatial image publishing on Xiaohongshu.

🔗 Related reading: vivo S60 Review: A Phone That Encourages Less Thinking, More Doing

Xiaomi 17T Series to Launch in Mainland China

Yesterday, Xiaomi Group President Lu Weibing announced that the Xiaomi 17T series will launch in Mainland China, expected to go on sale in early June.

Lu Weibing noted that Xiaomi first introduced the T series in 2019 with the “Xiaomi 9T”; over the past seven years, seven generations of T-series devices have directly competed against Apple’s iPhone and Samsung’s Galaxy S series.

According to his statement, the Xiaomi T series has amassed over 15 million users globally, covering Europe, Southeast Asia, Latin America, the Middle East, Japan, and over 100 countries and regions.

Recently, Xiaomi has already launched the Xiaomi 17T series in Hong Kong/Taiwan and Japan. The new lineup includes two variants: the 17T and 17T Pro. The former features a 6.59-inch flat display, Dimensity 8500-Ultra chipset, and Leica triple-camera system; the latter adopts the Dimensity 9500, a 6.83-inch flat display, and Leica triple-camera setup.

Base Kit at CNY 8,999: ZeroZero Tech Launches Haifu AQUA—Bringing Flying Cameras to Water Sports

ZeroZero Tech unveiled the Haifu AQUA, positioned as a fully waterproof flying camera designed specifically for water sports.

As the latest addition to the Haifu family, the AQUA weighs under 250 g, features IP67 dust/water resistance, positive buoyancy design, and surface takeoff/landing capability—ideal for hydrofoil boarding, wakeboarding, kiteboarding, kayaking, sailing, stand-up paddleboarding, and jet skiing.

The AQUA withstands winds up to Level 7 (up to 40 km/h), features a hydrophobic lens, anti-fog technology, and corrosion-resistant construction; its imaging system uses a 1/1.28-inch CMOS sensor capable of recording 4K video at up to 100fps, and includes over 15 intelligent tracking modes—including dedicated modes for SUP, kayaking, and hydrofoil boards.

World wakeboarding champion Dominik Gührs stated that the AQUA enables him to capture water sports footage solo, eliminating the need for photographers or complex equipment. The product is now available via domestic e-commerce platforms and offline channels, with the base kit priced at CNY 8,999 and the Fly-Freely kit at CNY 9,999.

Honor WIN Turbo Launched Starting at CNY 2,699

Honor WIN Turbo was launched yesterday, with a recommended retail price starting at CNY 3,299 and an introductory offer price starting at CNY 2,699. Available configurations include 12GB + 256GB, 12GB + 512GB, and 16GB + 512GB, priced at CNY 2,699, CNY 2,999, and CNY 3,599 respectively during the launch period.

Targeting large battery life and gaming scenarios, the device offers three color options: “Quick Squad Up,” “Target Win,” and “No Fear of Blue.” It supports IP68, IP69, and IP69K dust/water resistance ratings. The front features a 6.79-inch 1.5K LTPS OLED flat display with 8,000 nits peak brightness and 3,840Hz “zero-risk” dimming.

Powered by the Dimensity 8500 Racing Edition chip, the Honor WIN Turbo offers up to 16GB RAM and 512GB storage, houses a 10,000mAh Qinghai Lake battery, and supports 80W fast charging and 27W reverse charging.

Claude Code Launches Dynamic Workflow Preview

Anthropic announced the preview release of “Dynamic Workflow” in Claude Code’s CLI, Desktop, and VS Code plugins—designed for long-term development tasks such as cross-service codebase bug tracing, large-scale refactoring, framework migration, and pre-launch adversarial testing.

This mechanism transforms the previously single-threaded programming Agent into an orchestrated workflow: the main Agent writes orchestration scripts and delegates subtasks to dozens or hundreds of parallel sub-Agents;

Before merging code, the system schedules independent sub-Agents for multi-perspective adversarial testing and introduces two rounds of Reviewer Agents for code review. To support continuous operation, Dynamic Workflow incorporates incremental auto-save and checkpoint resume capabilities.

StepFun Open-Sources Step 3.7 Flash

StepFun released and open-sourced its next-generation Flash large model, Step 3.7 Flash, positioned as a foundational model for production-grade Agents, coding, web search, and multimodal workflows.

The model architecture comprises a 196B-parameter language backbone and a 1.8B-parameter visual Transformer, totaling ~198B parameters, with ~11B activated parameters, supporting 256K context length and three inference tiers.

Officially, Step 3.7 Flash can autonomously select, crop, and re-read localized information within UI interfaces or charts, and proactively retrieves and verifies information when uncertainty arises. Performance-wise, it achieves 67.1% on ClawEval-1.1 autonomous task execution, 79.2% on SimpleVQA Search, 95.3% on V* Python, and 56.3% on SWE-Bench Pro.

Liquid AI Open-Sources Edge MoE Model LFM2.5

Liquid AI released and open-sourced its edge-side Mixture-of-Experts model, LFM2.5-8B-A1B. With a total of 8B parameters, only 1B are activated per inference, targeting devices including smartphones, PCs, and robots—supporting local fine-tuning and offline operation on personal devices or a single consumer-grade GPU.

This update shifts LFM2.5-8B-A1B toward a “think-first, respond-later” reasoning paradigm and expands context capacity from 32K to 128K. Liquid AI also expanded the vocabulary to 128K to improve non-English text processing efficiency; according to their data, Thai and Hindi processing efficiency improved by 238.2% and 120.4%, respectively, while Arabic improved by 38.8%.

The team also implemented targeted optimizations for common small-model issues like cyclic reasoning and hallucination. Liquid AI states that targeted reinforcement learning makes the model more likely to admit ignorance beyond its knowledge boundary, increasing the no-hallucination rate from 7.46% in the previous generation to 63.47%.

Xiaomi Open-Sources Controllable Video Audio Model ControlFoley

Xiaomi’s official tech account released and open-sourced the multimodal controllable video audio generation framework ControlFoley, targeting automatic video dubbing scenarios. It extends “matching sound to visuals” to multi-condition control: creators can simultaneously input video, text instructions, and reference audio to specify desired sound content and timbre style.

ControlFoley primarily addresses three control challenges:

• When text and visual semantics conflict, the model prioritizes the user’s textual intent;
• When reference audio is provided, the model preserves the reference timbre without compromising audio-visual synchronization;
• When input condition combinations vary, unified multimodal representation and random modal dropout training ensure stable performance across TV2A, TC-V2A, and AC-V2A tasks.

Xiaomi’s experimental results show ControlFoley outperforms open-source baselines across multiple datasets in semantic alignment, temporal synchronization, and sound quality.

🤗 Hugging Face: huggingface.co/YJX-Xiaomi/ControlFoley

Tencent Hunyuan Releases Agent Memory Plugin Hy-Memory

Tencent Hunyuan launched Hy-Memory, a memory plugin designed for long-term collaborative agents like OpenClaw, aiming to resolve memory fragmentation, attention dilution, and outdated preference overwrite issues commonly encountered in multi-day, multi-session agent operations.

• Hy-Memory adopts a 6-layer memory architecture, decomposing memory into factual, profile, mental model, and forward-intent layers;
• It integrates a dual-system mechanism: System 1 extracts immediate facts and session summaries in milliseconds, while System 2 runs background processes to distill user mental models and knowledge networks;
• The system uses *supersedes* pointers to chain old and new memories into an evolution trail, ensuring retrieval results retain both the latest conclusions and preference evolution paths.

In evaluations, Tencent reports Hy-Memory scored 85.2 on LongMemEval and leads in preference retention, temporal reasoning, and knowledge updating metrics.

Performance-wise, its write speed is 8× faster than Graphiti, memory entry count is ~1/3 of mem0 and ~1/4 of Graphiti, and token consumption in long-context processing is reduced by 35%. Hy-Memory offers Lite, Pro, and Ultra configurations, defaulting to the Chroma local embedded vector database.

Starting Next Month, Refusing to Pay Fares After Violating Ticket Rules Will Result in Ticket Purchase Restrictions

CCTV News cited China State Railway Group, announcing that the revised *Railway Passenger Transport Regulations of China State Railway Group Co., Ltd.* will take effect on June 1, introducing for the first time ticket purchase restrictions for passengers who violate riding rules and refuse to pay supplementary fares.

Violations covered include traveling without a ticket, using altered, forged, or tampered travel documents, mismatch between ticket, ID, and passenger, occupying higher-class seats with lower-class tickets, and holding discounted or preferential tickets without valid eligibility documentation or failing to meet discount conditions.

When such violations occur and the passenger refuses to pay the fare supplement, railway staff will inform them on-site about the violation and subsequent ticket purchase restrictions. When restricted passengers attempt to purchase tickets again, railway operators will notify them of the restriction reason via 12306 or ticket windows. After paying the due fare plus a 50% surcharge, passengers can automatically lift the restriction online or offline;

After fare payment, passengers may still obtain electronic railway invoices within 180 days from the original trip’s completion date. The new regulations also adjust student ticket discount zones, refund policies for connecting tickets, and lost-item delivery rules.

Xiaohongshu Video Now Supports Dolby Atmos

Yesterday, Dolby Laboratories and Xiaohongshu jointly announced that Xiaohongshu has launched Dolby Atmos content sharing functionality. Following the introduction of Dolby Vision to Xiaohongshu in 2024, this new audio dimension allows creators to publish videos featuring both Dolby Vision and Dolby Atmos on compatible devices.

Xiaohongshu users can preview, share, and enjoy Dolby Atmos content via iPhones or Android phones supporting Dolby Atmos through the Xiaohongshu mobile app; they can also access and share/view content via Safari on Mac or iPad through the Xiaohongshu web platform.

Mixue Launches Mint Series New Products

Mixue announced the launch of its new Mint Series yesterday, emphasizing refreshing summer flavors. The lineup includes Mint Ice Cream, Mint Crispy Mini Sundae, Mint Milk Green Tea, and Snow King Snow-Topped Coffee (Mint Flavor).

According to official descriptions, Mint Milk Green Tea combines mint flavor, jasmine green tea, and creamy notes; Mint Ice Cream comes in Mint and Double Swirl variants; the Mint Crispy Mini Sundae consists of mint ice cream, chocolate crisp coating, and Oreo cookie crumbles.

Haoxianglai Removes Controversial NFC Juice

According to Jiemian News, following CCTV’s exposure of misleading labeling practices in fruit juice products, a series of juice drinks under the brand Haoxianglai drew public attention. The program pointed out that the product prominently displays “100% Juice” and “NFC (Not From Concentrate)” on its packaging, yet water and concentrated juice rank first and second in the ingredient list—indicating a mismatch between the label claims and actual composition.

A spokesperson for Wanchen Group, Haoxianglai’s parent company, responded that the product contains labeling inaccuracies, which may lead consumers to mistakenly believe it is pure NFC juice. The company has already pulled the product from shelves and plans to redesign the packaging to eliminate such ambiguity. The report also cited past controversies involving brands like Beibingyang and Master Kong over discrepancies between juice content and labeling.

✨ It’s the weekend!

One Fun Thing | Nintendo Made a Bizarre Game That Flips Through Your Photo Album

As reported by *The Verge*, Nintendo launched the mobile game *Pictonico* this week. It resembles a lightweight version of *WarioWare*: each round features 10 micro-games lasting only a few seconds, with on-screen prompts like “chomp,” requiring players to instantly interpret and execute actions—such as grabbing a mouth to make it chew food.

What makes it intriguing is that the game customizes gameplay using photos from your phone’s gallery. *The Verge* noted that players encounter absurd scenarios—plucking feathers, licking lollipops, peeling bananas—where their own photos are inserted into these mini-game scenes.

It doesn’t demand dedicated playtime; rather, it functions as a fragmented toy blending personal photo albums, split-second reactions, and Nintendo’s signature surreal humor. It transforms your photo library from something merely “viewed” into raw material for playful, interactive mini-games.

What to Watch This Weekend | *The Mandalorian & Grogu*

*Star Wars: The Mandalorian & Grogu* is the first theatrical film in Disney’s *The Mandalorian* TV series franchise, set chronologically after Season 3. Directed by Jon Favreau and co-written by Favreau, Dave Filoni, and Noah Kroll, the film stars Pedro Pascal.

The story centers on bounty hunter Din Djarin and his Force-sensitive ward Grogu—a non-biological father-son duo—who undertake a top-secret mission critical to the fate of the New Republic amid galactic instability following the Empire’s collapse.

The film continues the series’ core narrative theme of “father-son bonding,” advancing the plot through a game-like “level-up and boss-fight” structure. Its scope remains relatively contained, avoiding an expansive depiction of large-scale conflict between the New Republic and Imperial remnants.

As of May 29, 2026, the film holds a 7.4/10 rating on Douban (based on 27,407 reviews), outperforming 80% of action films and 83% of sci-fi films.

Book Buying Without Reading Guide | *Corporate Sickness: From Individual Struggles to Organizational Performance*

Author Xi Fanjun brings over a decade of experience working in major tech enterprises, specializing in strategic communications and technology governance. He now focuses on organizational structure and public narrative research.

Approaching the topic from a sociological perspective, the book draws on theoretical frameworks from Weber, Foucault, Bauman, and Bourdieu to systematically dissect the structural roots behind contemporary workplace phenomena such as “involution,” “lying flat,” and the “age-35 crisis.”

Its central argument posits that individual anxiety and burnout stem not from insufficient effort, but from the inevitable consequences of modern organizational alienation. Xi introduces original concepts—including “liquid humans,” “safe humans,” “performative governance,” and “the prison of language”—to deconstruct how large corporations reshape human behavior and erode meaning through processes, performance metrics, and rhetorical tactics.

The book argues that when companies evolve into self-sustaining systems, mid-level managers become “loyal hostages,” while senior employees face heightened marginalization risks—the recurring metaphor throughout the text being: “Process-correct, no survivors.”

Game Recommendation | Steam Freebie *Drift86*—Perfect for Weekend Drifting Practice

Independent developer RewindApp’s arcade-style drifting racer *Drift86* is currently available for free on Steam. Originally priced at ¥11 in China, the game can now be claimed for free and added permanently to your library.

*Drift86* emphasizes lightweight drifting mechanics, featuring over 40 vehicles and more than 30 tracks, supporting both single-player and multiplayer modes. Inspired by *Initial D*, the game pairs Eurobeat music with arcade-style cornering rhythms, focusing on quick entry into drifts, scoring, and route optimization loops.